Context-Aware Dynamic Reconfiguration of Mobile Patient Monitoring Systems

A mobile monitoring system for ambulant patients typically monitors bio-signals of a patient, processes and distributes these signals to formal caregivers. Often the monitoring service is long lasting, ranging from hours, days, to even months. The tasks involved in this end-to-end service are distributed over a networked ICT infrastructure and an optimal assignment may be found using task assignment techniques. To accommodate for the network and system resource dynamics, a configured system based on the optimal assignment over time may require a task redistribution. This capability is made possible using principles from autonomic computing where the operational environment is monitored, and based on significant operational changes a new optimal task assignment is computed, a reconfiguration plan is constructed and executed. In order for this approach to be effective, this control loop itself needs to be efficient, fast and minimally disruptive. In this paper we propose a computational scheme to estimate the quality of this control loop, the so-called reconfiguration cost. To this end, we propose an OSGi based task execution and control environment, a simple reconfiguration planning scheme based on the OSGi service bundle lifecycle, and a computational model to estimate the reconfiguration costs. Initial results using a prototype implementation are reported.